The primary aim of the proposed research program is to develop new psychometric methods for drug abuse, AIDS, and policy research. Methods to be developed include correct missing data methodologies, robust methodologies, distribution-free methods and a variety of other latent variable modeling methods applicable to the analysis of real drug abuse data, such as badly distributed nonnormal variables. These methods are then applied to import drug abuse datasets. In order to disseminate the results at various levels of technical detail, nontechnical papers and practical improvements in methodology are also developed. Tutorials to improve methodology in drug abuse research are also given, and drug abuse and health researchers and government agencies are advised on the methodology research results and its implications.